Using Electronic Monitoring to Document Snapper Discards and Validate Catch effort Data New Zealand Fisheries Assessment Report 2016/57 M.J. Pria, J.P. Pierre, H. McElderry, M. Beck, ISSN 1179-5352 (online) ISBN 978-1-77665-410-9 (online) October 2016
Requests for further copies should be directed to: Publications Logistics Officer Ministry for Primary Industries PO Box 2526 WELLINGTON 6140 Email: brand@mpi.govt.nz Telephone: 0800 00 83 33 Facsimile: 04 894 0300 This publication is also available on the Ministry for Primary Industries websites at: http://www.mpi.govt.nz/news resources/publications.aspx http://fs.fish.govt.nz go to Document library/research reports Crown Copyright Ministry for Primary Industries
Contents 1.0 Introduction... 3 1.1 Goal and Objectives... 3 1.2 Scope... 4 2.0 Materials and Methods... 4 2.1 Project Partners and Roles... 5 2.2 Field Operations... 5 2.2.1 Individual Vessel Monitoring Plans... 6 2.2.2 Electronic Monitoring System... 6 2.2.3 Onboard Methodologies and Camera Views... 7 2.3 EM Data Review and Analysis... 9 2.3.1 Sensor and Image Data Interpretation... 9 2.3.2 Imagery data quality... 10 2.3.3 Imagery Data Review Time and Analysis Ratio... 11 2.4 Reporting... 12 2.5 Catch effort Comparisons... 12 3.0 Results... 13 3.1 Data Collection Performance... 13 3.2 Fishing Location and Snapper Catch... 16 3.3 Snapper Discards... 17 3.3.1 Snapper discard characterization... 17 3.3.2 Snapper discard EM and catch effort comparisons... 24 3.3.3 Discarding methods... 26 3.4 Unidentified fish and other discards... 27 3.5 Imagery Data Review Time... 27 4.0 Discussion... 30 4.1 Data Collection Performance... 30 4.2 EM Onboard Methodologies... 31 4.3 Catch Effort Return Comparisons... 33 4.4 Other Considerations... 34 Data Collection Models... 34 Other Information Needs... 34 4.5 Next Steps... 35 5.0 References... 36
Appendix A: Addressing post retention discards through catch effort comparisons An example from the BC Groundfish Hook and Line fishery.... 38
Executive summary Pria, M.J.; Pierre, J.P.; McElderry, H.; Beck, M. (2016). Using Electronic Monitoring to Document Snapper Discards and Validate Catch effort Data. New Zealand Fisheries Assessment Report 2016/57. 38 p. The snapper (Pagrus auratus) fishery in quota management area one (SNA 1) is New Zealand s most valuable inshore finfish fishery. When the sustainability and management controls of the SNA 1 fishery were reviewed in 2013, wastage within the commercial fleet was identified by stakeholders as an area of concern. Of particular concern was the lack of information relating to how much snapper below the minimum legal size (MLS) was being returned to the sea. The Ministry for Primary Industries (MPI) contracted the services of Archipelago Marine Research Ltd. to examine the feasibility of electronic monitoring (EM) technology as a viable monitoring tool for estimating the amount of snapper returned to the sea and validating the estimates on the vessel catch effort data. EM systems were installed on five vessels. Vessel operators were asked to batch sub-mls snapper into specific bins in camera view and discard the snapper exclusively at pre-agreed discard points. EM imagery data was reviewed for catch sorting operations and bin fullness was recorded to estimate total undersize snapper discard weight per tow. Vessel operators were also required to record an estimate of snapper discard weight in their catch effort returns for comparison with EM estimates. EM data were collected for a total of 101 trips and 1187 tows across all five vessels, of which 392 were randomly selected for review to estimate sub-mls snapper discards. Overall, sub-mls snapper discards were observed in 59% of the reviewed tows. Snapper discards were less than 15% of the total snapper captured in 68% of the 162 tows. Catch effort estimates of snapper discards at the individual tow level were strongly correlated with EM estimates for all vessels (r ranged from 0.78 to 0.96) but underestimated discards. Underestimation varied by vessel, with Vessels 2, 3, 4, and 5 underestimating by 20% to 30% and Vessel 1 underestimating by 70%, relative to EM estimates. EM sensor data was captured for 90% of the reported tows as compared with catch effort return data. Missed tows were caused by technical issues and skippers forgetting to power on the EM system. Imagery data were successfully collected for 70% of the 392 tows randomly selected for review. Imagery collection success was affected by implementation challenges early in the trial and, to a lesser degree, technical issues. This trial was successful at developing a basic methodology for equipment set-up and catch-handling requirements, while helping to establish data review protocols for estimating snapper discards. It also provided initial verification that catch effort data tend to match EM snapper discards in situations where sub-mls snapper are binned before being returned to the sea. More broadly, this project builds on the past decade of pilot studies of EM in New Zealand inshore fisheries, demonstrating the efficacy and potential of EM in meeting a diversity of monitoring objectives. The results from this trial indicate that an EM programme would effectively assist MPI and SNA 1 Commercial efforts to develop accurate catch monitoring in this fishery. However, it is imperative that vessel operators follow the catch handling protocols developed in order to obtain reliable EM sub-mls snapper discard estimates. Deviating from the protocols (e.g., discarding outside of the control points, mixing snapper with other species, etc.) contributes to imprecision and bias in the EM estimates. Ministry for Primary Industries Using EM to Document Snapper Discards 1
Furthermore, not having a systematic way of handling catch, especially when dealing with large catch volumes, could also make it difficult for skippers to keep track of large amounts of sub-mls snapper discards. The key next step is to maintain effort in EM data collection and port infrastructure. Other next steps need to focus on developing a holistic EM programme design that will meet the information needs of the SNA 1 fishery, while identifying an optimal balance between data quality, data turnaround, and cost. 2 Using EM to Document Snapper Discards Ministry for Primary Industries
1.0 INTRODUCTION The snapper (Pagrus auratus) fishery in quota management area one (SNA 1) is New Zealand s most valuable inshore finfish fishery and it is utilised by customary Māori, recreational, and commercial sectors (MPI, 2013). Snapper is managed under the New Zealand quota management system (QMS). Current regulations that apply to the commercial take of snapper in SNA 1 require that snapper larger than the minimum legal size (MLS) of 25 cm must be retained and landed while any undersized specimens (sub-mls) must be discarded (MPI, 2013). When the sustainability and management controls of the SNA 1 fishery were reviewed in 2013, wastage within the commercial fleet was identified by stakeholders as an area of concern (MPI 2013). Of particular concern were the lack of information in regarding the amount of snapper discards in the inshore trawl fleet. To help address these issues, the commercial industry, represented by the SNA 1 Commercial group, worked with the Ministry for Primary Industries (MPI) to improve the fishery-dependent data for the inshore trawl snapper fleet fishing in SNA 1. This involved two components: Improving fisher-reported data by having skippers record the weight of undersized snapper discards for each tow in their catch effort returns. Increasing monitoring through the use of electronic monitoring technology. The intent was to support the monitoring coverage targets set for the fishery of 25% of the effort of the SNA 1 fleet by 1 December, 2013, 50% coverage by October 1, 2014, and 100% coverage by 1 October, 2015 (MPI and SNA 1 Commercial, 2013). Electronic monitoring (EM) is being trialled or is operational in many fisheries around the world. It is implemented as an alternative and/or a complement to onboard human observers (Mangi et al. 2015; McElderry, 2008). The EM systems, developed by Archipelago Marine Research Ltd. (Archipelago), incorporate technology that has proven successful in monitoring fishing activity and collecting fisheries-related data across a range of applications for more than a decade (McElderry, 2008). Each EM system consists of a centralized computer control centre that records data on an array of sensors and cameras, and provides information on key aspects of the fishing operations such as vessel location, vessel speed, equipment activity and catch sorting and stowing. The Ministry for Primary Industries (MPI) contracted the services of Archipelago to examine the feasibility of EM technology as a viable monitoring tool for estimating snapper discards and validating catch effort data. 1.1 Goal and Objectives The overall goal of the project was to examine the efficacy of EM for verifying catch effort data of weight estimates of sub-mls snapper discards 1. Specific objectives of the project can be summarised as: Summarise EM equipment performance; Develop EM onboard methodologies specific to this fishery, including equipment set-up, catch-handling requirements, and data processing protocols to estimate snapper discard weights per tow from EM imagery; and Summarize results of a comparison of the EM and catch effort data. 1 Note that discarded snapper were not verified as being sub-mls as part of this trial. See Section 4.2 for a discussion of this topic. Ministry for Primary Industries Using EM to Document Snapper Discards 3
1.2 Scope The project included the development of individual vessel monitoring plans (IVMP) that specified the EM installation requirements for the vessel and involved in-season feedback to MPI and vessel owners on actions to improve data quality. Archipelago was also required to build local capacity by training local field and data technicians and provide overall project advice on all aspects of the project. 2.0 MATERIALS AND METHODS As part of the trial, MPI nominated ten vessels for coverage with EM systems. Of these ten vessels, five were assigned to Archipelago. The remaining five vessels participated in a parallel trial led by Trident Systems (Middleton et al., 2016). The ten vessels were selected based on their catch profiles, fishing companies and geographical location. In the previous fishing year they also accounted for the majority of the trawl effort targeting snapper in SNA 1 and over 70% of the landed SNA 1 catch. The results from the Trident trial are reported elsewhere. Starting in January 2014, MPI and SNA 1 Commercial led a series of planning meetings that included representatives from Archipelago and Trident Systems. During these meetings, a general trial methodology was developed: All catch was to be brought aboard for sorting; All sub-mls snapper was to be batched in specific bins before being discarded; All discarding was to occur in camera view from one or two discard control points, to be agreed on with the vessel owner. A memorandum of understanding (MOU) was developed between SNA 1 Commercial and MPI to set the terms of engagement of the vessels participating in the trial (MPI and SNA 1 Commercial, 2013). The trial methodology described below was based on the terms described in the MOU. In particular, the MOU specified that: EM imagery data were to be encrypted; Imagery recording was to occur for the entire trip, (i.e., port-to-port and 24 hours a day, seven days a week); At least two cameras were to be deployed (general overview and detailed view of discard points); Fish were to be discarded from no more than two points within the unobstructed view of the camera; Sub MLS snapper discards were to be batched to allow quantification for each tow; The purpose of imagery recording was to capture all catch sorting and discarding events; and All discards were to be reported in kilograms (kg). 4 Using EM to Document Snapper Discards Ministry for Primary Industries
2.1 Project Partners and Roles The project sponsor (MPI) was responsible for overall programme direction, and served as the main contact with vessel owners. This work included securing vessels to participate in the programme, ensuring vessels were able to carry EM equipment, and leading outreach with industry. SNA 1 Commercial, a working group of commercial fishers, quota owners, and Licensed Fish Receivers (LFRs) in SNA 1, provided advice and support on programme direction. Archipelago was responsible for the day-to-day coordination of the project, providing advice on the programme design, training the local field and data technicians, overall data analysis and reporting. Ongoing maintenance of the EM systems in the port areas was provided by local field technicians subcontracted by Archipelago. The local field technician was responsible for providing service to the vessels when required, including data retrievals, troubleshooting, and moving or replacing sensors or cameras. Archipelago subcontracted Johanna Pierre Environmental Consulting Ltd. to provide programme advice, and to complete a portion of the data processing locally in New Zealand. 2.2 Field Operations Archipelago staff and the local field technicians installed the EM systems in late February, 2014. Technicians met with each vessel representative (i.e., vessel owner or fleet manager depending on the vessel) to discuss the installation and use of the EM system. Data collection began in late March or early April, 2014, depending on when each vessel representative signed the MOU. Data collection was planned for six months and ended at the end of August 2014. EM systems were removed in October 2014. Vessels ranged in length from 15.4 to 18.2 meters. EM systems were installed on five vessels; two in Auckland and three in Whangarei. Subsequently, vessels were serviced in these ports as well as in Totara North because one vessel relocated partway through the trial. EM technicians visited the vessel within one week of each vessel s first EM trip to verify the EM system setup. EM technicians then visited the vessels on a regular basis to retrieve data and check the EM system s functionality. Additional services were provided as needed to address any issues or concerns about the monitoring equipment. Data retrieval schedules varied by vessel depending on the amount of data collected. The original plan was to visit the vessels monthly and collect data on 500 GB hard drives, both measures intended to reduce programme costs. However, initial data collection showed that the hard drives could be filled in less than a month. Hard drives were then changed to 1 TB capacity and data retrievals were scheduled on a biweekly basis for vessels with particularly high levels of fishing activity. This ensured data collection continuity. To protect their privacy, participating vessels were identified using a vessel ID in this report, rather than the vessel name. Ministry for Primary Industries Using EM to Document Snapper Discards 5
2.2.1 Individual Vessel Monitoring Plans Meeting the data requirements relied on understanding the combination of catch-handling protocols and EM system configurations. The primary data needs that drove the standards for catch handling and equipment installation were the need to verify that discarding had occurred within the designated control points, and the requirement to estimate the amount of sub-mls snapper by observing what was in the bins being discarded and assessing how full they were. EM reviewers did not estimate the length of snapper being discarded but assumed they were sub-mls snapper. Catch-handling and equipment installation standards were documented in the IVMP. The IVMP is a communications tool designed to help vessel representatives/skippers, EM field technicians, EM data reviewers, and project coordination staff to understand their roles for a successful implementation. Each IVMP used a combination of narrative and images to document the key points related to vesselspecific EM installation and operation: General vessel information; Skipper responsibilities; EM system configuration: o o General description of the type of data being recorded, and Location and objective of each EM system component (including camera views); Catch-handling protocols: o o Catch sorting, and Discard control points; Diagram of the vessel; and Software configuration specifications (for EM technician reference). 2.2.2 Electronic Monitoring System The EM systems were designed and manufactured by Archipelago in Victoria, BC, Canada, specifically for the purpose of monitoring and collecting fishing-activity data at sea. The EM system consisted of an EM Observe TM v4.5 control centre with an array of digital closed circuit television cameras, a GPS receiver, a hydraulic pressure sensor, and a rotational sensor (Figure 1). The EM Record TM operating software, installed on the control centre, collected high-frequency sensor data every ten seconds throughout the entire trip, and recorded imagery data. Skippers were to power EM systems on at departure from port and only power them off upon return to port. The EM systems operated independently and were set to record imagery using a geo-fencing feature in the EM Record software. At the time of installation, the EM technician configured the EM system to record imagery whenever the vessel was inside of the SNA 1 quota management area (QMA). Imagery data recording was triggered by geo-fencing, and began immediately inside the SNA 1 QMA, and would stop ten minutes after the vessel went into a different QMA or entered the port area. Imagery and sensor data were encrypted and stored digitally on a removable hard drive. 6 Using EM to Document Snapper Discards Ministry for Primary Industries
Figure 1: Schematic of a standard EM Observe v4.5 system used during this study. 2.2.3 Onboard Methodologies and Camera Views Catch Handling Vessels had slightly different methods for how the catch was brought on board, sorted and discarded or stowed. However, crews generally sorted catch on deck. Retained catch was batched into bins and stowed in the hold or in containers on the deck and sub-mls snapper was batched in bins and discarded over the side rails or stern usually once all catch was sorted. Other discarded catch such as porcupine fish (Allomycterus jaculiferus), skates and rays, and kingfish (Seriola lalandi) was generally discarded individually during sorting. The main exception to the catch handling protocol described above was Vessel 1, on which the codend was sometimes emptied down the central hatch into the hold. In this situation, catch sorting took place below deck out of camera view. Discards, including sub-mls snapper, were batched in bins, brought up through the centre hatch, and discarded over the rails. Camera Views Camera placement varied by vessel depending on the vessel configuration and catch-handling practices. Three cameras were used per vessel, with the exception of one vessel for which vessel size and layout required four cameras. There were two general camera configurations. The first consisted of a general deck overview, and close up views of the catch sorting areas and discard control points (Figure 2). This configuration was used on Vessels 1, 3 and 5. Ministry for Primary Industries Using EM to Document Snapper Discards 7
Figure 2: Example of the first general camera configuration used. Clockwise from top left: view of the trawl net coming in on the stern; deck overview; starboard catch sorting and discard control point area; and port catch sorting and discard control point areas. The fourth camera was used to provide a close up view of the port slurry bin in case it was used to sort catch. The second general configuration, for Vessels 2 and 4, did not have a deck overview but instead had a camera view covering the area off the stern of the vessel to observe the trawl net coming onboard as well as two cameras covering the deck where catch was sorted, discarded and stowed (Figure 3). Figure 3: Example of the second general camera configuration used. Clockwise from top left: port side catch stowage area; starboard catch sorting and discard control point areas; and view of the trawl net coming in on the stern. 8 Using EM to Document Snapper Discards Ministry for Primary Industries
All cameras were mounted to existing superstructure on the fishing vessels. Vessel representatives provided guidance as to where the cameras could be mounted and were least likely to be damaged or have views obscured during regular vessel operation. 2.3 EM Data Review and Analysis 2.3.1 Sensor and Image Data Interpretation After the EM data were retrieved from a vessel they were passed to a group of EM reviewers, who were responsible for data review and analysis. The EM data sets were reviewed using the Archipelago EM Interpret TM Pro software, a specialized software package designed to help the reviewer quickly process, evaluate, and report on fishing activity. The EM Interpret Pro software integrates thousands of imagery, sensor, and GPS records into a single synchronized profile, and presents it along a common timeline. This allows reviewers to quickly follow cruise tracks, review fishing events times and locations, and verify catch handling. Key events, comments and observations can be saved as annotations, created by the reviewer and saved along with the data set for future reference. All information is then stored in a standard database format for easy reference, analysis, or secondary processing. Trip Tow Catch Sorting Figure 4: Example of typical sensor data signature for a tow. Sensor information is shown on a timeline with speed in green, winch rotation in blue, and hydraulic pressure in red. A typical trawl shot signature (highlighted in orange) consisted of high winch rotation counts and hydraulic pressure. Vessel speed and hydraulic pressure would remain relatively constant while the trawl was towed. During gear haul-back (highlighted in purple) the vessel would slow down followed by high winch rotation counts and hydraulic pressure. Reviewers examined all EM data to check that sensor and imagery data recording was complete, to identify individual fishing trips, and to identify the time and location of all fishing events. Each fishing event was comprised of four time periods: trawl shot; towing; gear haul back; and catch sorting (Figure 4). For selected tows, imagery data were reviewed from the end of the tow to the end of catch stowage. For the purpose of this report we use the term tow to refer to the entire fishing event, as opposed to only the period when the trawl net was towed. Reviewers randomly selected thirty percent of the tows identified in the sensor data per trip, with a minimum of one tow per trip for imagery review. If the selected tows were unusable or had no imagery data (see below for definitions), replacement tows were randomly selected from the pool of remaining usable tows in the trip. Three EM reviewers were assigned to process the imagery data for this trial. Imagery review consisted of watching catch sorting operations which were defined as the time from the end of the tow (to observe if all catch was brought on deck), until all catch on deck was seen stowed above or below deck and Ministry for Primary Industries Using EM to Document Snapper Discards 9
discarding seemed to be complete. In cases where catch was sorted out of camera view (below deck), the imagery reviewer would watch until either the fish were brought into camera view for release or until the trawl doors came up at the end of the following tow (note that reviewers did not watch video collected between catch sorting events). To estimate sub-mls snapper discards, reviewers documented the type and number of discarded snapper bins as well as the fullness of each. Three types of bins were observed (Table 1). Table 1: Snapper discard bins types, dimensions and estimated weight when full. Estimated Dimensions Bin Type Weight (kg) (L x W x D in cm) Standard Iki 20 59.7 36.2 20.5 Sanford Standard 30 67 39 27 Non-standard Unknown Unknown To estimate the snapper weight discarded, reviewers visually estimated fullness of each bin to the nearest 25% increment. A trace category was reserved for bins that were less than 10% full to avoid inflating very small discard amounts but was never required because snapper were discarded individually whenever there were only a few in a tow. If the fish was not discarded using a bin, snapper discards were recorded as individual piece counts. These were converted to weight using a mean piece weight of 0.3 kg. Estimated weights for standard bins and snapper discards were obtained from commonly assumed weights used by industry and approved by MPI for the purposes of this trial. For snapper discards in non-standard bins, the snapper discarded weight was visually estimated. Nine of a total of 291 bins observed were noted to contain a proportion of other species. These were flagged as mixed species bins but treated in the analysis as if the composition was 100% snapper (see Section 3.3.3 for a discussion of the potential bias). The imagery reviewer also recorded discards of other species using the same methodology as snapper discards (i.e., bin counts with estimated fullness or piece counts of individual fish discards). Incidents where discarding was seen to occur outside the control points documented in the IVMP were noted. No attempts were made to obtain length measurements for discarded snapper or estimate retained weights for any species. 2.3.2 Imagery data quality The quality of the imagery was assessed for tows selected for review based on whether the imagery data were complete and whether reviewers were able to reliably detect snapper discards, not only due to the clarity of the images but also based on catch handling being conducive to meaningful observations. Imagery quality for each reviewed tow was averaged across all cameras according to the following guidelines (see Figure 5 for examples): 10 Using EM to Document Snapper Discards Ministry for Primary Industries
High: Imagery was very clear and the viewer had a good view of fishing activities. Focus was good, light levels are high and all discard activity was easily seen. Medium: Imagery was acceptable, but there may be some difficulty assessing discards. Slight blurring or slightly darker conditions hamper, but did not impede analysis. Low: Imagery was difficult to assess. Some camera views may not have been available. Imagery was somewhat blurred or lighting was inadequate. Some factors such as crew standing between the catch and the camera for extended period of time may have also occurred. Unusable: Imagery was available but poorly resolved or obstructed such that snapper discards and fishing activity could not be reliably discerned. There was no imagery from one or more cameras (but not all) and those camera views were critical to record snapper discards. No Data: No imagery available from any camera for the entire tow. Figure 5: Examples of image data quality. Top left high quality, top right medium quality due to water on camera, bottom left medium quality due to low light, bottom right unusable quality due to colour saturation on the camera. 2.3.3 Imagery Data Review Time and Analysis Ratio Reviewers were able to adjust the imagery playback speed from frame-by-frame to up to 16 times realtime speed as well as stop and go back at specific frames as necessary depending on the activity on deck. For example, imagery of crew sorting catch was generally reviewed at two to four times real-time speed but slowed down and reviewed frame-by-frame at specific times to confirm discards. The time required to review the catch sorting for each tow was recorded to calculate the amount of imagery review effort for estimating snapper discards. An analysis ratio was calculated by dividing the hours of real-time imagery by the time taken to review it. Ministry for Primary Industries Using EM to Document Snapper Discards 11
2.4 Reporting Throughout the trial, results for each vessel were provided to the vessel owners and to MPI via a monthly summary report within six weeks after month-end. The monthly summaries included an update on system performance, trips and tows reviewed, data collection and completeness, observed snapper discards in bins and as individuals, observations that did not meet the IVMP, and recommendations to increase data quality. Catch effort data were not available in a timely manner in-season so EM vs. catch effort comparisons were not included in the monthly in-season reports. 2.5 Catch effort Comparisons Skipper reporting requirements were modified starting March 1, 2014, such that skippers were to report a greenweight estimate of the sub-mls snapper that was caught and returned to the sea under the species code SNX (to differentiate it from retained snapper, recorded under the code SNA). Skippers did not record sub-mls snapper previous to this. The catch effort estimates of sub-mls discards (SNX) were compared with EM estimates for the same tows. For the purpose of this trial, the EM data were considered the standard. Skipper catch effort data were provided by MPI for all vessels participating in this trial up to the end of the data collection period (August 31, 2014). EM and catch effort trips and tows were paired using date and times. This allowed identification of trips and tows that were not captured by the EM system due to the system being powered off and, conversely, verification that all trips and tows captured by EM were reported in the catch effort data. 12 Using EM to Document Snapper Discards Ministry for Primary Industries
3.0 RESULTS 3.1 Data Collection Performance During this trial, EM data were collected from 101 trips and 1187 tows (Table 2). Fishing activity levels varied by vessel. Vessel 3 represented 45% of the total tows captured. The mean number of tows per trip ranged from 8 (Vessel 5) to 19 (Vessel 3). The catch effort returns recorded 105 trips and 1313 tows for the same time period. Overall, 90% of the tows reported by the catch effort were identified in the EM data and the tow data collection was 95% or higher for three vessels. Catch effort records were complete to the extent that no additional trips or tows were identified in the EM data. Table 2: Summary of data collected. Trips and tows identified in the EM sensor data and reported in the catch effort returns. Percent of tows reported in the catch effort returns and captured in the EM sensor data. Vessel EM EM Catch effort Catch effort Tows with EM ID Trips Tows Trips Tows Sensor Data 1 15 146 19 208 70% 2 18 187 18 188 99% 3 28 533 28 559 95% 4 24 198 24 230 86% 5 16 123 16 128 96% Totals 101 1 187 105 1 313 90% Technical issues and having the EM system powered off resulted in 10% of the tows not being captured by EM sensor data (Table 3). Technical issues included problems with the control centre on two occasions on Vessel 1, and the EM system was powered off on Vessels 1 and 3 for most of one trip each when they were experiencing camera connectivity issues. Vessels 4 and 5, but in particular Vessel 4, would often power the EM system off overnight or when the engine was off for extended periods and would sometimes forget to turn it on when they resumed fishing operations. Vessel 3 had vessel power issues that caused them to keep the EM system off, missing six tows, which was resolved when the vessel installed a battery dedicated to the EM system. Table 3: Summary of tows without EM sensor data by issue. Number of Percentage of Tows Percentage of Issue Category Tows Affected not Reviewed Total Tows Technical Issue Control centre 45 36% 3% Camera connectivity- EM off 37 29% 3% Subtotal 82 65% 6% Other Issues No Power 44 35% 3% Total 126 10% Ministry for Primary Industries Using EM to Document Snapper Discards 13
A total of 392 (33% of the captured tows) were randomly selected for imagery review from the tows captured by EM sensor data. Of those, 274 (70%) were successfully reviewed while 118 could not be reviewed either because no imagery data were collected or the imagery was unusable (Table 4). Of the tows selected that could not be reviewed, 27 were replaced by randomly reselecting other tows from the same trip. Table 4: Summary of imagery data usability for tows selected for review. Tows Selected Tows Tows Not Vessel ID for Review Reviewed Reviewed 1 56 46 10 2 58 55 3 3 170 77 93 4 65 59 6 5 43 37 6 Totals 392 274 118 Percent total 70% 30% Table 5 provides a summary of the reasons why some of the selected tows could not be reviewed. EM deployments always involve a bedding-in period when systems and processes are being established, personnel are being trained, and stakeholders are becoming familiar with the technology and the specific objectives of the trial. Such issues are typically resolved as deployments progress, especially in the medium or longer term. In this project, implementation issues resulted in 20% of the selected tows not being reviewed. Technical issues, mainly problems with camera connectivity, resulted in 10% of the selected tows not being reviewed. Camera connectivity issues occurred on three vessels and caused one or more cameras to not record imagery for a period of time (Table 5). Table 5: Reasons for not being able to review the 118 tows. Number of Percentage of Tows Percentage of Implementation Tows Affected not Reviewed Selected tows Recording Configuration 47 40% 12% Camera Positioning 22 19% 6% Power disruption 6 5% 2% Disc Space 3 3% 1% Subtotal 78 66% 20% Technical Issue Camera connectivity 37 31% 9% Camera colour saturation 3 3% 1% Subtotal 40 34% 10% Total 118 14 Using EM to Document Snapper Discards Ministry for Primary Industries
Imagery data quality was high for 53% and medium for 46% of the reviewed tows (Table 6). The most common reason for medium quality was dark images during night tows. In these instances, the images still allowed identification of snapper discards. Most hauls occurred during the night. Assuming night time to be between 19:00 and 06:00 hours, approximately 56% of the total EM tows (1187) would be considered night hauls and only 5% of the tows would be considered to have occurred within full daylight hours (Figure 6). There were only three tows with low quality imagery. Snapper discards were difficult to observe in these tows due to 1) extremely low light levels combined with crew positioning, 2) the catch being sorted in an area partly obstructed from camera view, and 3) missing one camera view but with the remaining cameras providing a view of the catch sorting and discarding. Table 6: Imagery quality ratings for reviewed tows. Vessel ID High Medium Low 1 2 3 4 5 Totals Percentage 35 21 41 30 18 145 53% 11 34 35 27 19 126 46% 1 2 3 1% Figure 6: Percent of EM tows with both start and end times occurring at night (56%), tows with only the end occurring at night (25%), tows with only the start occurring at night (13%), and tows with both start and end times occurring during the day (5%). Label values represent the number of tows. Night time was assumed to be between 19:00 and 06:00 hours. Ministry for Primary Industries Using EM to Document Snapper Discards 15
3.2 Fishing Location and Snapper Catch During the trial, the participating vessels fished exclusively in SNA 1 (Figure 7). Fishing activity was observed mostly in the north Hauraki Gulf area (Statistical Areas 005 and 006) by the vessels fishing out of Auckland. The vessels operating out of Whangarei and Totara North fished mostly along the coastal regions of Northland (Statistical Areas 002 and 003). Figure 7 shows a spatial summary of snapper fishing effort, reported as the number of tows by start location. Retained snapper catch is shown by tow start location in Figure 8. Figure 7: Chart of New Zealand s northeast North Island showing EM fishing activity by total number of tows within 10 10 km grids for all five participating vessels combined. Tows were assigned to grids according to their start location. 16 Using EM to Document Snapper Discards Ministry for Primary Industries
Figure 8: Total retained snapper (kg) recorded in the catch effort returns for the 274 viewed tows across all five participating vessels. Each grid represents a 10 10 km area and the values in each grid represent the number of tows. 3.3 Snapper Discards 3.3.1 Snapper discard characterization Sub-MLS snapper discards were observed in 162 (59%) of the reviewed tows (Figure 9). Sub-MLS snapper discards were a small proportion of the total snapper captured for the reviewed tows (total snapper was calculated as the sum of retained snapper reported in the catch effort returns and discarded snapper observed in the EM data) (Figure 9). Sub-MLS snapper discards were less than 15% of the total snapper captured in 229 (84%) of the tows reviewed and exceeded 25% in 12 tows (4%). Some total snapper captured amounts per tow were very small which tended to inflate the percentages of discarded amounts. The amount of sub-mls snapper discards varied widely among tows as indicated by the estimates of the coefficient of variation 2 (CV) in Table 7 and Figure 10. Furthermore, the frequency and volume of sub-mls snapper discards varied between vessels (Table 7). Maximum discard weight per vessel varied from 35 kg/tow for Vessel 2 to 270 kg/tow for Vessel 1 (Figure 10). In general, the frequency distributions of snapper discards per tow by vessel are left-skewed, (i.e., there was a large proportion 2 We use CV (standard deviation/mean) to characterize to the variation in the raw observations (Snedecor & Cochran 1967), as opposed to the alternative usage of CV in which it is used as a synonym for the relative error (RSE) of the mean. Ministry for Primary Industries Using EM to Document Snapper Discards 17
of relatively small discard volumes per tow for each vessel and a few tows with large discard volumes) but the shape of the distribution varied widely among vessels (Figure 10). Figure 9: EM snapper discards as a percentage of total snapper (retained + discarded snapper) for all 274 reviewed tows across all five vessels. 18 Using EM to Document Snapper Discards Ministry for Primary Industries
Table 7: Characterization of snapper catches and discards for the reviewed tows, by vessel. Number of tows reviewed, number and percentage of tows with snapper discards, total snapper discarded weight (kg), mean (CV%) discarded snapper weight (kg) for tows with snapper discards, and mean (CV%) retained snapper weight (kg) (as reported in the catch effort returns) for tows with snapper discards. CV%=Coefficient of Variation (standard deviation/mean) expressed as a percentage. Mean Mean Snapper Snapper Tows with Total Snapper Discards/tow Retained/tow Vessel Tows Snapper Discards (kg ) (CV%) (kg) (CV%) (kg) ID Reviewed Discards EM EM EM FLOG 1 46 36 (78%) 2 390.0 66.3 (105%) 362.1 (82%) 2 55 27 (49%) 144.9 5.4 (132%) 171.8 (68%) 3 77 75 (97%) 2 267.1 30.2 (88%) 270.5 (95%) 4 59 11 (19%) 114.2 10.4 (121%) 200.0 (96%) 5 37 13 (35%) 187.2 14.4 (149%) 334.6 (82%) Total 274 162 (59%) 5 103.4 31.5 3 274.8 2 Figure 11, Figure 12, and Figure 13 respectively show the spatial distribution of total sub-mls snapper discards, mean snapper discards per tow as reported by EM and catch effort returns, and the proportion of snapper catch discarded across all vessels. The majority of sub-mls snapper discards occurred in the northern Hauraki Gulf area in Statistical Areas 005 and 006. Some total snapper captured amounts per tow were very small resulting in inflated percentages for discarded amounts. 3 Note that overall variation for the total estimate (not shown) will depend on stratification. Ministry for Primary Industries Using EM to Document Snapper Discards 19
Number of Tows 0 5 10 15 20 25 30 Vessel 1 Vessel 2 Number of Tows 0 5 10 15 20 25 30 0 20 60 100 140 180 220 260 300 0 20 60 100 140 180 220 260 300 EM Snapper Weight (kg) EM Snapper Weight (kg) Number of Tows 0 5 10 15 20 25 30 Vessel 3 Number of Tows 0 5 10 15 20 25 30 Vessel 4 0 20 60 100 140 180 220 260 300 EM Snapper Weight (kg) 0 20 60 100 140 180 220 260 300 EM Snapper Weight (kg) Number of Tows 0 5 10 15 20 25 30 Vessel 5 0 20 60 100 140 180 220 260 300 EM Snapper Weight (kg) Figure 10: Histograms of EM snapper discards (kg) by individual vessel for tows with snapper discards. 20 Using EM to Document Snapper Discards Ministry for Primary Industries
Figure 11: EM (top) and catch effort (bottom) total snapper discards (kg) for all 274 reviewed tows across all five participating vessels. Each grid represents a 10 10 km area, and the values in each grid represent the number of tows. Tows are assigned to grid cells using start location. Ministry for Primary Industries Using EM to Document Snapper Discards 21
Figure 12: EM (top) and catch effort (bottom) average snapper discards (kg) for all 274 viewed tows across all five participating vessels. Each grid represents a 10 10 km area, and the values in each grid represent the number of tows. Tows are assigned to grid cells using start location. 22 Using EM to Document Snapper Discards Ministry for Primary Industries
Figure 13: Sub-MLS snapper discards as a percentage of total snapper catch for all 274 reviewed tows across all five vessels for each 10 10 km area grid. EM proportions= sum of EM discards by grid / (EM discards + retained snapper). Catch effort proportions= sum of catch effort discards by grid / (catch effort discard + retained snapper). Ministry for Primary Industries Using EM to Document Snapper Discards 23
3.3.2 Snapper discard EM and catch effort comparisons There was agreement in presence/absence of snapper discards in EM and catch effort returns for 255 of the 274 reviewed tows (93%) (Table 8). Of these, 102 tows (37%) recorded an absence of snapper discards. For the remaining 19 tows (7%) either EM reported snapper discards and the catch effort did not, or vice versa. Estimated discarded amounts (either from EM or catch effort) for these tows were small and ranged from 0.2 to 9 kg. Table 8: Snapper discard matrix for the number of tows when both EM and catch effort returns recorded snapper discards, when no snapper discards were recorded by either, and when only EM or catch effort returns recorded snapper discards. Catch effort Catch effort Snapper No Snapper Discards Discards Totals EM Snapper Discards 153 10 163 EM No Snapper Discards 9 102 111 Total 162 112 274 The EM and catch effort estimates were correlated for each vessel, but varied in degree with r varying from r=0.78 for Vessel 1, 0.82 for Vessel 3 to over 0.94 for Vessels 2, 4, and 5 (Figure 14) Catch effort returns for all vessels tended to underestimate sub-mls snapper discards. Catch effort estimates tended to be 70% to 80% of the EM estimates for Vessels 2, 3, 4, and 5. The underestimation bias was particularly strong in Vessel 1, where catch effort estimates tended to be 30% of the EM estimates overall and estimates were very heavily underestimated for larger values (i.e., EM estimates larger than 100 kg). Note that although the results suggest that the catch effort data for this vessel did not record discards over 100 kg, an examination of the catch effort data for the entire data collection period revealed that this was not the case. The catch effort data did include some tows (not selected for review) that reported discards over 100 kg. A closer examination of the tows with EM estimates over 100 kg revealed that these tows had high or medium image quality, so video quality was unlikely to be a contributing factor. However, deviation from the standard catch handling protocols designed for the trial may be affecting the estimates for these tows. In two tows from Vessel 1, snapper was mixed in with other species in some of the discard bins. This deviation from the catch handling protocols may be resulting in an overestimation of EM sub-mls snapper discard estimates or conversely the snapper discarded with other species may have not been properly accounted for in the catch effort returns. In two additional tows for Vessel 1, catch was discarded outside of the discard control points. In the remaining six tows there were no notable circumstances to explain the large discrepancies between the catch effort and the EM estimates. 24 Using EM to Document Snapper Discards Ministry for Primary Industries
A B C Figure 14: Catch effort vs. EM snapper discards (kg) for Vessel 1 (top left), Vessel 2 (top right), Vessel 3 (middle left), Vessel 4 (middle right), and Vessel 5 (bottom) for tows with reported sub-mls snapper discards. The dashed line represents the zero intercept and slope of one. Ministry for Primary Industries Using EM to Document Snapper Discards 25
Large differences where catch effort data overestimated snapper discards compared with EM were also investigated. Data points marked as A and B in Figure 14 are both tows where discards occurred outside of the control points (at the stern and through the scuppers respectively). However, we could not detect any obvious reasons for discrepancy shown as C in the vessel 3 observations. 3.3.3 Discarding methods EM reviewers collected information on the onboard methodologies used to discard snapper and whenever discarding occurred outside of the discard control points. The majority of snapper was binned into standard containers (either iki or Sanford bins) before being discarded (Table 9). However 158 snapper were discarded individually across 61 tows. The latter practice was mostly observed on Vessel 3. Vessel 1 was the only vessel that used non-standard containers and that discarded other species in the same bin as snapper. As mentioned in Section 2.3.1, there were nine bins out of 291 bins observed where a proportion of other species were mixed in with snapper. These were treated as being all sub-mls snapper because snapper appeared to dominate in most of these bins and it was not considered feasible to develop a consistent protocol for estimating bin-specific proportions among reviewers. The total estimate of sub- MLS snapper discards from these five bins was 218 kg, or 4% of the total EM sub-mls snapper discards in the study. Table 9: Summary of the number of tows with EM snapper discards by discard method. The tow counts do not add to the total tows as more than one discard method may have been used during catch sorting for a single tow. Vessel Tows with Standard Individual Mixed Non-Standard ID Snapper discards Bin Snapper Species Bin 1 36 30 8 5 2 2 27 21 10 0 0 3 75 71 30 0 0 4 11 7 7 0 0 5 13 11 6 0 0 Total 162 140 61 5 2 Fish were discarded outside of the control points in 5% of the reviewed tows (15 tows), (Table 10). Discards outside of control points were mainly done at the stern of the vessel whereas the discard control points were along the side of the vessel at the rails. In addition to stern discarding, crew on Vessel 1 swept fish off the deck or pushed discards towards the scupper (two tows) and on Vessel 4 bins were discarded at the rail but outside the control point (two tows). There is a possibility that not all discards were observed for the tows where catch were discarded outside of the control points. Observations regarding catch sorting and discarding and their impact in obtaining EM estimates were provided as feedback to vessel representatives in the in-season reports. 26 Using EM to Document Snapper Discards Ministry for Primary Industries
Table 10: Number of trips and tows in which discards were observed outside of control points. Vessel ID Trips with Discards Tows with Discards Outside Outside of Control Points of Control Points 1 4 6 2 1 1 3 1 1 4 6 7 5 0 0 Total 12 15 3.4 Unidentified fish and other discards The study recorded over 13 000 individual fish being discarded that were not identified as snapper (Table 11). Of these, 10 504 (80%) were identified to species. The main factor precluding identification was that fish were tossed overboard so rapidly during sorting that their body shape was reduced to only a blurred outline. This mostly occurred at night although not exclusively. The identified discards of individual fish comprised mainly porcupine fish (Allomycterus jaculiferus), skates and rays, and kingfish (Seriola lalandi). Table 11: Summary of EM unidentified fish and non-snapper discard by fish count. Unidentified Other Total Other Vessel ID Fish Species Discards 1 160 531 691 2 285 1 645 1 930 3 556 2 536 3 092 4 1 362 2 257 3 619 5 307 3535 3 842 Totals 2 670 10 504 13 174 Additionally, there were 25 bins of unidentified fish discarded on Vessel 1 across 16 tows. There were two scenarios in which these bins were observed. The first occurred whenever catch was sorted below deck and non-snapper discards would get combined in bins and brought up to deck for discarding. In this scenario it was difficult to identify all the species in the bin. The second scenario was when crew sorted catch close to the stern and the bin was not brought into camera view. In this scenario the reviewers were able to observe the discard bin in the deck overview but it was not possible to identify the species within. 3.5 Imagery Data Review Time A total of 274 tows were reviewed. The catch sorting time associated with these tows totalled 334 hours of imagery. The review was completed in 172 hours (Table 12) leading to an analysis ratio of 0.51. In other words, it averaged about 30 minutes to review one hour of imagery although the ratio for any given hour was highly variable. The amount of imagery data reviewed per tow varied significantly between tows from a little as 20 minutes to as much as 25.7 hours, with a mean of 1.2 hours. There were two main reasons for this variation. One was the amount and composition of catch in the tow and the second was whether the catch was sorted and stowed promptly. The latter reason could have a large impact. For example, the Ministry for Primary Industries Using EM to Document Snapper Discards 27